Computing surface roughness of individual cells of digital elevation models via multiscale analysis

In this paper, a modification of the algorithm proposed by Ahmad Fadzil et al. (2011) for surface roughness computation from digital elevation models (DEMs) via multiscale analysis is presented. The new algorithm takes into account that the three predominant physiographic features of terrains (mountains, basins and piedmont slopes) have distinct curvature region distributions and hence, distinct roughness characteristics. To this end, the surface roughness of individual cells of DEMs is computed by identifying the curvature region distribution and roughness characteristics of each individual mountain, basin and piedmont slope region. The modified algorithm allows for the localisation and quantification curvature regions over varying scales for specific regions, providing a more appropriate surface roughness parameter.

► A previous algorithm is extended to perform the computation of surface roughness of individual cells of DEMs.
► The DEM terrain is segmented into the three predominant physiographic features; mountains, basins and piedmont slopes.
► The mask of cells modified, which are the curvature regions, is computed from the generation of multiscale DEMs.
► The curvature regions are analysed to compute surface roughness of individual cells of each distinct individual landform.
► The modified algorithm allows for the localisation of surface roughness computation.